FABBS Undergraduate Research Excellence Award

Megan Maxwell, Washington University in St. Louis, Department of Psychological & Brain Sciences

“Evidence that Neighborhood Threat and Brain Volume Mediate the Relationship Between Neighborhood Poverty and Children’s Psychopathology”

Megan Maxwell, Washington University in St. Louis, Department of Psychological and Brain Sciences

Abstract

Higher rates of internalizing/externalizing behaviors have been observed in children from more impoverished neighborhoods while increased crime rates or altered brain structure may contribute to this relationship. This project examined the extent to which neighborhood poverty (NP) relates to children’s mental health outcomes independent of household socioeconomic status, and whether neighborhood threat, brain volume, and the structural integrity of white matter mediated this relationship. Data from the Adolescent Brain Cognitive Development study were obtained for 8,623 9-10 year old children; NP was measured using nine variables from the Area Deprivation Index and neighborhood threat included parent-/child-reported feelings of safety and objective crime rates. Amygdala and dorsolateral PFC volumes were obtained by MRI and FA values of the cingulum and uncinate were obtained by DTI. Psychopathological symptoms were assessed by the Child Behavior Checklist. Generalized linear models and structural equation modeling tested for evidence of a mediation. Increased NP was significantly associated with increased externalizing symptoms, and parent/child reported feelings of safety and left hemisphere amygdala volume mediated this relationship. These results highlight environmental and neurological markers that may increase risk for later psychopathology that can better inform interventions designed to reduce the burden of mental health.

Introduction/Background

The environment in which you are raised has a powerful effect on one’s future health outcomes, particularly the effects of family environment and neighborhood context on mental health. For example, early poverty has been associated with increased risk for mental illnesses, reduced cognitive function, and lower performance in both school and on tests (Taylor et al., 2020). In addition, an accumulation of neighborhood adversity has been associated with an increase in allostatic load—the ‘wear and tear’ on a body—in adulthood (Brody et al., 2014). However, previous literature has largely only considered an individual’s household socioeconomic status (SES) as a predictor of these important developmental outcomes, and there may be other aspects of a neighborhood not fully captured by individual household poverty that can still impact children’s health outcomes. The aim of this project was to investigate the extent to which a wider definition of poverty that includes neighborhood characteristics in addition to household poverty explains outcomes of internalizing and externalizing disorder symptomology in children. We are also interested in how specific brain structures and tracts are impacted by the link between neighborhood poverty and psychopathology, specifically the amygdala, dorsolateral prefrontal cortex (dlPFC), uncinate fasciculus, and bilateral cingulum. Disruptions in the volumes of these structures and abnormal fractional anisotropy values of the tracts have been shown to be indicative of the detrimental effects of neighborhood poverty and may contribute to increased risk for mental health problems.

Internalizing behaviors include problems such as noncompliance. Consistent patterns of increased internalizing symptoms have been observed in withdrawal, depression, and anxiety whereas externalizing problems are characterized by aggression, hyperactivity, and more impoverished neighborhoods, with factors such as chronic stress mediating the effect of neighborhood SES on internalizing type behaviors (Kim et al., 2013). Additionally, correlates of neighborhood poverty such as increased toxin levels or perception of safety may mediate the relationship between poverty and externalizing behaviors (Nevin, 2007; Goldner et al., 2016). One potential explanation underlying the associations between poverty and developmental outcomes is that impaired neural development may be a central mediating pathway, as changes in certain brain structures and tracts have been implicated in both conditions of poverty and psychopathology. Internalizing and externalizing type disorders are most commonly related to reduced amygdala and dlPFC volumes, while abnormal mean diffusivity of the uncinate fasciculus and bilateral cingulum is linked to internalizing and abnormal fractional anisotropy was related to externalizing disorders (Andre et al., 2020). Another mechanism underlying the association between neighborhood poverty and children’s psychopathology may be exposure to crime. The neighborhood is a primary context for children’s development, embedded with additional sources of stress that are not captured at the household level. Previous literature has found that exposure to neighborhood characteristics such as crime may augment children’s feelings of fear and distress, contributing to an increased risk for anxiety and depressive symptoms (Rabinowitz et al., 2016). Regarding externalizing type problems, research has found that engaging in more problem behaviors could confirm a more frequent perception of danger (Goldner et al., 2016).

Methods

Data already collected from the national longitudinal Adolescent Brain Cognitive Development (ABCD) study was used to assess the relationship between neighborhood poverty and psychopathology in 8,623 nine and ten year old children. Neighborhood poverty was measured using an existing composite measure consisting of nine census-tract variables from the Area Deprivation Index, which were assessed through residential history questionnaires administered through ABCD (Taylor et al., 2020). The nine variables include percent of labor force >16 unemployed, percentage of families below the poverty level, percentage of population below 138% of the poverty threshold, percent of single parent households, percentage of occupied housing units without a motor vehicle, percentage of population aged >/= 25 at least a high school diploma, percentage of homeowners, median family income, and income disparity. Individual household SES was calculated using combined income of the primary caretaker and any additional household members. Internalizing and externalizing disorder symptoms were gauged using the parent-report Child Behavior Checklist in which multiple subscales were combined to form a “broad” composite measure of internalizing and externalizing symptoms; internalizing behaviors consisted of anxiety, depression, and worry subscales whereas externalizing behaviors were grouped by the hyperactivity, non-compliance, and attention problem subscales.

This study looked at amygdala and dlPFC volumes and the structural integrity of the uncinate and cingulum tracts. The volumes of brain structures and the structural integrity of white matter tracts were obtained through magnetic resonance imaging and diffusion tensor imaging, respectively. Neighborhood threat was created as a construct that encompassed perception of crime and objective crime rates. Perception of crime involved parent- and child-reported answers to the ABCD Parent (or Youth) Neighborhood Safety/Crime Survey that assessed feelings of safety through questions such as “My neighborhood is safe from crime.” Objective crime rates were measured by the Uniform Crime Reports database in which participants’ zip codes were used to generate crime rates at their specific county level that included total adult offenses such as drug sale and possession, adult violent crimes (burglary, murder, rape), and DUI’s.

Generalized linear models following a Baron and Kenny approach (Baron & Kenny, 1986) first tested for a direct relationship between neighborhood poverty and internalizing/externalizing disorder symptoms such that they are significantly related even after taking parent income into account. Then, evidence of a mediation of brain variables and/or neighborhood threat mediating the relationship between neighborhood poverty and children’s psychopathology was tested for. If this was supported, structural equation modeling was then used to more explicitly examine indirect pathways.

Results

We found that increased neighborhood poverty was significantly associated with increased externalizing disorder symptoms after accounting for parent income in the model (p = 0.04). On the other hand, increased neighborhood poverty was significantly associated with decreased internalizing disorder symptoms (p = 0.04). Regarding a mediation, we followed a Baron and Kenny approach (Baron & Kenny, 1986) and attempted to establish separate significant pathways between the predictor variable of neighborhood poverty, each of the mediator variables, and the outcome variables of internalizing and externalizing problems. The results indicated that parent- and child-reported feelings of safety significantly mediated the relationship between neighborhood poverty and externalizing behaviors, while left hemisphere amygdala volume was significantly associated with only externalizing disorder symptoms.

Parent-reported feelings of safety were significantly associated with neighborhood poverty (p < 0.001 ) and with externalizing disorder symptoms (p < 0.001); however, when the indirect effect of parent-reported feelings of safety (p < 0.001 ) was considered, the direct effect of neighborhood poverty on externalizing behaviors was no longer significant—evidence for a full mediation. Lastly, child-reported feelings of safety were significantly associated with neighborhood poverty (p < 0.001 ) and with externalizing disorder symptoms (p < 0.001 ); however; when the indirect effect of child-reported feelings of safety (p < 0.001 ) was considered, the direct effect of neighborhood poverty on externalizing behaviors was no longer significant—evidence for a full mediation. Objective crime rates, dlPFC and right hemisphere amygdala volume, and the fractional anisotropy values of the uncinate and cingulum were not found to be significant mediators.

Discussion

Our hypotheses were partially confirmed, particularly that neighborhood poverty would be related to children’s mental health outcomes above and beyond individual household SES and that brain variables and neighborhood threat would be implicated to an extent as mediators of that relationship. However, we were surprised to find that neighborhood poverty was negatively associated with internalizing problems such that increased neighborhood poverty was significantly related to decreased internalizing behaviors. This may be because children from more impoverished neighborhoods employ suppression as a means of coping with stress in the neighborhood as suggested by McCoy, Raver, & Roy (2016)’s findings. Alternatively, the presence of social or emotional support can potentially moderate the otherwise negative impact of neighborhood poverty on children’s mental health. For example, a study investigating the relationship between neighborhood poverty and allostatic load in African-American youth found that the highest levels of allostatic load occurred for youth who lived in more impoverished neighborhoods, but only in the context of low emotional support (Brody et al., 2014).

The results also indicated that perception of neighborhood safety—not objective crime— was a better indicator of neighborhood threat mediating the relationship between neighborhood poverty and children’s externalizing problems. This may be explained by the scope of the Uniform Crime Reports measure, as this database generates a crime index for participants’ county based on their home addresses. Because county level crime estimates do not directly align at the neighborhood level, it may be that these values do not accurately reflect the perceived neighborhood threat children experienced in their own neighborhood context.

The findings in regard to the left hemisphere amygdala volume may also be related to previous research that has found evidence for amygdala lateralization. There is a hemisphere- specific difference with the left hemisphere being activated more often than the right while engaging in emotional processing, hence this may help explain the significant findings for the left hemisphere but not the right (Baas, Aleman, & Kahn, 2004). On the other hand, dlPFC volume was not a significant mediator despite being significantly related to neighborhood poverty and an independent predictor of externalizing disorder symptoms. Similarly, none of the white matter tracts mediated the relationship between neighborhood poverty and children’s externalizing behaviors. It may be that there are alternative brain structure pathways that we did not examine that better explain this relationship.

One potential limitation of the current study is its cross-sectional design because it only allows us to look at associations between the variables and not to establish causality. Ideally, a longitudinal version of this study can be carried out in the future once ABCD gathers data across multiple time point assessments. Doing so would allow us to more confidently make predictions about increased neighborhood poverty potentially contributing to risk for poor mental health outcomes and the mechanisms by which this occurs. Additionally, there may be other mediator or moderator variables not accounted for that play a role in the relationship between neighborhood poverty and children’s psychopathology, such as other brain structures or toxin levels.

Impact Statement

There are many different factors that contribute to the development of mental health challenges later in life, and exploring these possibilities during a formative time such as adolescence is paramount for understanding how to prevent them. Our approach in this study is novel because we are accounting for parent income to determine if neighborhood poverty explains children’s mental health outcomes above and beyond household SES. Understanding altered trajectories of children’s brain development and considering the broader neighborhood context as a predictor of psychopathology is important for identifying neurological and environmental markers associated with later mental health problems. As a result, this can inform the development of more holistic interventions designed to reduce the burden of mental illness.